Automatiseeritud kliendikontaktide analüüs finantstehnoloogia sektori näitel
Files
Date
2018
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Seoses infotehnoloogia arenguga tekib igapäevaselt enneolematu kogus andmeid, mille automaatne analüüsimine konkurentsieelise saavutamiseks on otsustava tähtsusega. Traditsioonilised andmekaeve meetodid on leidnud laialdaselt ärilisi rakendusi, kuid ei ole sobivad struktureerimata (näiteks tekstiliste) andmete puhul. Seevastu on valdav osa andmetest just struktureerimata kujul, mistõttu on iseäranis oluline luua lahendusi neist olulise teabe eraldamiseks. Käesolev magistritöö on praktilise loomuga ning selle eesmärk oli luua automatiseeritud tekstianalüüsi mudel, mida saab kasutada sissetulevate kliendipäringute efektiivseks prioriseerimiseks ning mõõtmiseks kasutades TransferWise Ltd. andmeid. Tulenevalt püstitatud eesmärgist teostas autor arvukalt eksperimente kasutades nii klassikalisi kui ka uudseid loomuliku keele töötluse meetodeid. Seejuures ei taganud antud ülesande puhul uudsed tehnoloogiad märgatavat paremust klassikaliste meetodite ees. Töö tulemusena valminud mudel on oluline nii ettevõttele kui ka selle klientidele – mudel võimaldab prioriseerida sissetulevaid päringuid vastavalt nende keerukusele ning pakilisusele, mis parandab kliendikogemust ning soodustab ettevõtte kasvu muutes operatsioonilisi protsesse efektiivsemaks. Peale praktilise väärtuse pakub käesolev töö ka ulatuslikku ülevaadet erinevatest loomuliku keele töötluse meetoditest, nende sobivusest ning nendega kaasnevatest võimalustest.
The rapid development of information technologies has brought along abnormal amounts of data being generated on a daily basis and the need to automatically analyse it to gain a competitive advantage. Traditional data mining techniques have been efficiently applied in a variety of commercial applications, yet they are only applicable on structured data. However, an overwhelming amount of existing data is in an unstructured (e.g. textual) form, hence it is crucial for companies to build solutions to automatically extract useful information from it. Given master’s thesis is with a practical nature and its purpose was to implement an automated text analysis model using data from TransferWise Ltd. that can be used to efficiently prioritise and measure incoming customer contacts. To achieve this, the author conducted numerous experiments via employing classical as well as novel natural language processing techniques. Apropos, employing novel methods did not ensure a noticeably better outcome. The established model is important for both the company as well as its customers since it can be used to prioritise incoming contacts based on their complexity or urgency. This ensures a convenient customer experience and is likely to accelerate growth by making operational procedures more efficient. Besides its practical value, given thesis also provides an extensive comparison of numerous natural language processing techniques, their suitability and opportunities.
The rapid development of information technologies has brought along abnormal amounts of data being generated on a daily basis and the need to automatically analyse it to gain a competitive advantage. Traditional data mining techniques have been efficiently applied in a variety of commercial applications, yet they are only applicable on structured data. However, an overwhelming amount of existing data is in an unstructured (e.g. textual) form, hence it is crucial for companies to build solutions to automatically extract useful information from it. Given master’s thesis is with a practical nature and its purpose was to implement an automated text analysis model using data from TransferWise Ltd. that can be used to efficiently prioritise and measure incoming customer contacts. To achieve this, the author conducted numerous experiments via employing classical as well as novel natural language processing techniques. Apropos, employing novel methods did not ensure a noticeably better outcome. The established model is important for both the company as well as its customers since it can be used to prioritise incoming contacts based on their complexity or urgency. This ensures a convenient customer experience and is likely to accelerate growth by making operational procedures more efficient. Besides its practical value, given thesis also provides an extensive comparison of numerous natural language processing techniques, their suitability and opportunities.